Implementation of angular_rate_error in roll_pitch_yawrate_thrust_controller.cpp does not match with the paper

[TaoYibo1866]

According to Control of complex maneuvers for a quadrotor UAV using geometric methods on SE(3) https://arxiv.org/pdf/1003.2005.pdf, the anglular rate error should be Eigen::Vector3d angular_rate_error = odometry_.angular_velocity - R.transpose() * R_des * angular_rate_des;. However, in https://github.com/ethz-asl/rotors_simulator/blob/master/rotors_control/src/library/roll_pitch_yawrate_thrust_controller.cpp line 111, the implementation is Eigen::Vector3d angular_rate_error = odometry_.angular_velocity - R_des.transpose() * R * angular_rate_des;. Is this a bug?

[TYXDG]

Hello, I have noticed the problem you mentioned. In addition, when calculating the expected angular acceleration in the computedesiredangularacc function, angular velocity is cross-multiplied by angular velocity, which is obviously an invalid operation. What do you think? Thank you very much!

*angular_acceleration = -1 * angle_error.cwiseProduct(normalized_attitude_gain_) - angular_rate_error.cwiseProduct(normalized_angular_rate_gain_) + odometry_.angular_velocity.cross(odometry_.angular_velocity);

[TaoYibo1866]

This cross-multiplication is valid, but always return zero. In this paper, there is

M = −kR * eR − kΩ * eΩ + Ω × JΩ

The code you mentioned just omitted J.

[scarlett-sun]

I also wonder why the controller for M omits the last part(the long expression) when it is easy to implement?

[barrosiuri]

Did you guys find out why did they omit the last part of the expression and why they are performing an operation that always returns zero (cross product with itself)?